A high school student from Bangalore applies machine learning to one of aerospace engineering's toughest challenges—and identifies optimal material combinations for next-generation spacecraft.
Arjun Vinayaka Doss, a student at National Academy for Learning in Bangalore, India, has completed research using artificial intelligence to optimize thermal protection systems for spacecraft—the critical materials that protect vehicles during atmospheric re-entry.
His paper, "AI-Enhanced Fibre-Reinforced Composites for Aerospace Thermal Protection Systems," applies machine learning to NASA's TPSX database to predict and rank composite materials for extreme heat environments.
When spacecraft re-enter Earth's atmosphere, they experience extreme temperatures that can exceed 2,000°C. The materials protecting these vehicles—called Thermal Protection Systems (TPS)—must balance two competing requirements:
- Insulation: Preventing heat from reaching the spacecraft interior
- Structural endurance: Withstanding extreme temperatures without failing
Traditional material testing is expensive and time-intensive. Arjun's research asks: can AI accelerate this process?
Working with his YRI mentor, Arjun developed an AI-driven framework that:
- Analyzed NASA's TPSX database of thermal protection materials
- Created a new metric called the Thermal Shielding Efficiency Index (TSEI) to evaluate combined performance
- Trained machine learning models (Random Forest and Multi-Layer Perceptron neural networks) to predict material performance
- Identified optimal hybrid combinations for next-generation spacecraft
| Model | Performance |
|---|---|
| Random Forest | R² = 0.56, MAE = 245 K |
| MLP Neural Network | Captured nonlinear relationships between thermal properties |
The models revealed that thermal conductivity is the strongest predictor of temperature endurance, followed by the TSEI metric.
Arjun's analysis identified an optimal hybrid design:
- LI-900 Silica Tiles: Best for insulation (TSEI = 128, max temp = 1,260 K)
- Carbon-Phenolic: Best for structural endurance (TSEI = 0.3, max temp = 2,200 K)
Combining these materials creates a system that balances high insulation with superior high-temperature performance—exactly what's needed for reusable spacecraft and hypersonic vehicles.
Arjun's research demonstrates how AI can:
- Reduce dependency on expensive high-temperature experimentation
- Rapidly evaluate new fiber-matrix combinations
- Provide data-driven insights for designing lightweight, reusable aerospace composites
The framework he developed is applicable beyond spacecraft—it can inform material selection for any high-temperature application, from hypersonic aircraft to industrial furnaces.
This isn't Arjun's first research achievement. He has also:
- Published in Springer Nature with research on nanotechnology
- Filed an Indian patent (Application No. 202541112136) for a photocatalytic system converting CO₂ to O₂ using BiVO₄ nanomaterials
His aerospace research represents a continuation of his work at the intersection of AI and materials science.
Before joining the YRI Fellowship, Arjun had strong interests in physics, materials science, and machine learning—but no clear path to conducting publishable research.
Through YRI, he worked with a mentor who helped him:
- Identify a meaningful research problem
- Access and analyze NASA's materials database
- Develop novel metrics for material evaluation
- Apply machine learning to aerospace engineering challenges
The result: original research that contributes to one of the most important problems in space exploration.
The YRI Fellowship pairs high school students with PhD-level mentors to conduct original research, publish papers, and build profiles that stand out. Learn more about the program or apply today.